Professional Documents
Culture Documents
www.springerlink.com/content/1738-494x(Print)/1976-3824(Online)
DOI 10.1007/s12206-015-1016-y
An investigation on the supersonic ejectors working with mixture of air and steam
Maziar Shafaee1,*, Mohsen Tavakol1, Rouzbeh Riazi1 and Navid Sharifi2
1
Faculty of New Sciences and Technologies, University of Tehran, 14395-1561, Iran
Departments of Aerospace Engineering, Amirkabir University of Technology, Tehran, 15875-4413, Iran
(Manuscript Received December 13, 2014; Revised May 23, 2015; Accepted June 21, 2015)
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Abstract
This study evaluated the performance of an ejector using two streams of fluids as suction flow. Three motive flow pressures were considered when investigating ejector performance; the suction flow pressure was assumed to be constant. The suction flow consisted of a
mixture of air and steam and the mass fraction of air in this mixture varied from 0 to 1. The ejector performance curves were analyzed for
different mass fractions of air. The results indicate that variation of the mass fraction of air in the suction flow mixture had a significant
effect on ejector performance. At all motive flow pressures, the ejector entertainment ratio increased as the mass fraction of air in the
suction flow increased. The results also show that the sensitivity of ejector performance to variation in the mass fraction of air in the suction flow decreases at higher motive flow pressures. An increase in motive flow pressure caused the transition from supersonic to subsonic flow to occur at higher ejector discharge pressures.
Keywords: Computational fluid dynamics; Entrainment ratio; Steam ejector; Supersonic flow; Working fluid
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
1. Introduction
Ejectors are designed to maximize the suction of secondary
fluid, thoroughly mix the primary and secondary fluids and,
depending on the design requirements, increase the secondary
flow pressure pumped to a region of high pressure. Ejectors
operate with compressible and incompressible fluids and have
applications in the refrigeration, aerospace, water desalination
and petroleum industries [1, 2].
In the refrigeration industry, ejectors have superseded mechanical compressors to compress refrigerant. Ejectors are a
main component of solar cooling systems [3-13] that use renewable energy technology. In desalination systems, ejectors
compress water vapor, which is known as Thermal vapor
compressing (TVC). One application of TVC is for multieffect desalination units. In desalination units a TVC returns
redundant steam to the evaporators, which have higher pressure levels [14].
Keenan et al. [15] initiated theoretical work on ejectors.
Their studies led to the first analytical model of ejectors. The
constant-area mixing model and constant-pressure mixing
model proposed by Keenan provided a framework for ejector
studies. DeFrate and Hoerl [16] improved the constantpressure mixing model by assuming ideal gas consumption
and by taking into account the effect of molecular weight.
*
Munday and Bagster [17] modified the Keenan model to consider chocking and aerodynamic throat theory. Huang et al.
[18] improved the previous 1D analytical model by considering the aerodynamic throat area to give a more accurate prediction of ejector performance. The model provided by Huang
et al. [18] was valid only in the double-choked mode. Chen et
al. [19] improved the Huang model to obtain more accurate
results in the single choked condition.
Analytical models cannot provide sufficient information
about the ejection process because of the complexity of flow
in ejectors. Improvements in computational power and methods have allowed scientists to better understand flow phenomena. Hedges and Hill [20] used Computational fluid dynamics
(CFD) to devise a finite-difference arrangement to simulate
flow inside an ejector. Riffat and Omer [21] studied the performance of a methanol-driven ejector using a commercial
CFD package. Sriveerakul et al. [22] used Fluent software to
simulate the flow inside a steam ejector. They validated their
numerical results using experimental test data. They used
steam as the working fluid by assuming perfect gas and isentropic flow in the ejector.
Hemidi et al. [23] used turbulence models to predict the
ejector entrainment ratio. Simulation results of different turbulence models and experimental data were compared to determine the most accurate models. They found that deviation
from the experimental data was less than 10% for the k-
model; however, this deviation was increased for the k--sst
turbulence model. Bartosiewicz et al. [24] evaluated six turbu-
4692
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
Suction Flow
NXP
SuperSonic
flow
Motive
Flow
SubSonic
flow
shock
wave
Mixing
Chamber
OUT FLOW
Constant
Area
Section
Diffuser
Fig. 1. Schematic of a conventional ejector with different flows and geometrical zones.
4693
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
(1)
CR =
pdis
.
psuc
(2)
In Eq. (1), mmot and msuc are the primary and secondary mass
flow rates, respectively. In Eq. (2), pdis and psuc are the pressure
values at discharge and the suction boundaries, respectively.
Every ejector has a specific performance curve that behaves
differently under different operating conditions. One important parameter is CR. Fig. 2 describes the effect of CR on ER
in the double-choked flow, single-choked flow and the reversed flow regions. If back pressure pdis is less than a certain
critical pressure p*dis, ER remains constant because of choking
of the secondary fluid in the aerodynamic throat (doublechoked flow condition). An increase in back pressure beyond
p*dis prevents the secondary fluid from reaching M = 1 and it
remains un-choked; only the primary stream is choked in the
primary nozzle (single-choked flow condition). A sudden drop
in the secondary mass flow rate can be calculated in this mode.
A further increase in ejector back pressure causes the flow to
return to the ejector and the suction chamber. This operational
condition is known as the reverse flow mode.
3. Model description
3.1 Numerical procedure
Commercial CFD software (Fluent 6.3) was used to study
flow behavior in an ejector. This program employs the finitevolume method for discretization. An implicit density-based
solver solves the nonlinear equations in the steady state condition. For all equations, the convective terms are discretized
using a second-order upwind scheme. The fluid flow inside
the ejector is considered to be axisymmetric [27].
Regarding the employment of boundary condition, it should
be noted that in the present study, since the velocity of the
flow entering and leaving the computational domain is relatively small compared with the supersonic flow condition of
the ejector; there would be no difference between employing
the boundary conditions of stagnation pressure and static pressure. Consequently, for two entering faces of primary nozzle
and secondary inlet stream, the pressure-inlet boundary condition was considered in this work. Also for the outlet of subsonic diffuser of the ejector, the pressure-outlet boundary condition was implemented. It is noteworthy that for all different
operating conditions which have been investigated in this
study, the abovementioned pressure type boundary conditions
were considered.
Since the flow in the ejector is high speed, heat transfer
through the walls is assumed to be negligible.
Single
Choked
mode
Reversed
flow
mode
ER =
Double Choked
mode
Compression ratio(CR)
Fig. 2. Ejector performance curve.
(3)
( r ui )
p t ij
( r uiu j ) = +
+
t
x j
xi x j
(4)
(rE)
( r u i E + u i p) =
+
t
xi
p
(k eff
)+
(u i t ij ) .
+
t xi
xi
x j
(5)
w = 1
i
(6)
i =1
wi =
mi
mtot
(7)
4694
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
Dimension (mm)
95
24
Table 2. The results of grid dependency study for pure steam as ejector
working fluid.
Fig. 3. Variation of Mach number along the centerline of ejector for
cases of pure steam (Black line) and pure air (Red line) as the secondary inlet stream.
calculations were also performed using other turbulence models (i.e., k- RNG, k- standard, k--sst, and k- standard),
however, their results have not been presented here for the
sake of brevity. It should be noted that, among these studied
turbulence models, the results of using turbulence model of k realizable for the case of pure steam were in good agreement
with related experimental results [22].
The case of pure air as the secondary inlet stream was investigated because it was assumed that the study on distribution of the air mass fraction in the flow field and degree of
mixing could become more feasible by increasing the amount
of air in the suction stream. Sharp variations (with high gradient) in the amount of Mach number, on the centerline of ejector, could be assumed as a measure to determine the regions
where strong or weak shocks might occur. It is noteworthy
that the results of k- realizable with pure air show the same
behavior as those of pure steam (see Fig. 3). However, as indicated in Fig. 3, the maximum difference for the amounts of
calculated Mach number between the cases of pure air and
pure steam is occurred in the range of around x~0.12 and
x~0.25.
Based on the obtained results from various turbulence models, the turbulence model of k- realizable was considered as
the suitable method (compared with other turbulence models)
for modeling the effects of variation in mass fraction of the
suction flow in the present study. To calculate flow near the
wall, the logarithmic standard wall function is used. It calculates flow velocity for near-wall nodes using empirical coefficients. The main governing equations of this model are expressed in Eqs. (8) and (9):
m
( r k ) + ( rk ui ) =
m + t
t
x
x j
sk
k
x + sk
j
(8)
( re ) + ( re ui ) =
m + t
t
x
x j
se
e
+ se
x j
(9)
Case
Mesh nodes
Entrainment ratio
18740
0.5273
26700
0.5456
39000
0.5456
4695
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
Table 3. The results of grid dependency study for mixture of air and
steam as ejector working fluids.
Case
Mesh nodes
Entrainment ratio
18740
0.440
26700
0.445
39000
0.445
the mixture of air and steam respectively) in terms of the influence of grid density on the calculated values of entrainment
ratio using the turbulence model of k- realizable.
As shown in Tables 2 and 3, by increase of the gird resolution from 18740 up to 27600 the amount of computed entrainment ratio increases. However, increasing the grid density
beyond the amount of 27600 would not influence the calculated entrainment ratios. Also the effect of grid resolution on
variation of Mach number along the centerline of ejector (for
the case of wair = 0.5) has been studied as shown in Fig. 5.
Here again, the results show that by increase of the grid density beyond a certain value, the results of variation of Mach
number do not change.
3.6 Validation of CFD results
The data from Sriveerakul et al. [22] was used to validate
the CFD results. The ejector used in their study was modeled
under similar operating conditions. The simulation incorporates vapor stream as the motive flow. A comparison of the
results listed in Table 4 shows that the numerical results are in
good agreement with the experimental results under all operat-
Table 4. Comparison between results of numerical modeling and the experimental data [22].
Operating conditions
Symbol
Entrainment ratio
Primary flow
Secondary flow
Saturated tempera- saturated temperature
ture (Motive flow)
(Suction flow)
Experiment
CFD results
(Present study)
Deviation from
experimental results
[22] (%)
0.53
0.5445
-1.88
37
33
Experimental
results [22]
CFD results
(Present study)
120
10
130
10
0.4
0.40251
-2.5
50
45
140
10
0.28
0.30355
3.57
65
60
4696
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
0.8
0.7
0.6
0.6
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
w=1
w = 0.67
w = 0.5
w = 0.33
w=0
0.7
ER
ER
0.8
w=1
w = 0.67
w = 0.5
w = 0.33
w=0
Fig. 7. Ejector performance curve for operating condition of , considering different values for mass fraction of air at suction (wair).
0.8
0.5
ER
ER
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
2
CR
Fig. 8. Ejector performance curve for operating condition of , considering different values for mass fraction of air at suction (wair).
a (w=1)
a (w=0)
b (w = 1)
b (w = 0)
g (w=1)
g (w=0)
0.7
0.6
0.8
0.6
Fig. 9. Ejector performance curve for operating condition of , considering different values for mass fraction of air at suction (wair).
w=1
w = 0.67
w = 0.5
w = 0.33
w=0
0.7
4
CR
CR
CR
flow of the ejector [29]. Figs. 7-9 have been integrated in Fig.
10 to demonstrate how the ER changes as mass fraction of air
changes at different motive flow pressures.
Fig. 10 shows that as the pressure of the motive fluid increased, the ER values remained constant for an extended
range of CR values, which further signifies that increasing the
pressures increases stability. Fig. 10 also show how ER varies
as the mass fraction of air varies. All performance curves indicate an increase in ER with an increase in the mass fraction of
air; however, an increase in ER upon an increase in the mass
fraction of air in the suction flow differed for different motive
flow pressures (Fig. 10).
The variation in ER as a function of variation in the mass
fraction of air in the suction flow for g is lower than for
a and b. In other words, the sensitivity of ejector performance
to the variation in mass fraction of air in the suction flow decreased at higher motive flow pressures (higher momentum of
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
5. Conclusions
The performance of an ejector using two gases as working
flows and the effect of variation of the mass fraction of air in
the suction flow was investigated. To the knowledge of the
authors, little study has been done on the influence of variation
in mass fraction of air in suction flow on ER of the ejector.
With the aid of numerical simulation, the conditions for two
streams of air and steam were analyzed assuming a perfect gas.
The initial numerical analysis for the ejector employed pure
steam as the working fluid. After the validation of the initial
case with experimental data from related studies [22], the effect of different mixtures of air and steam for the suction flow
were investigated. Ejector performance was analyzed for all
feasible values for the mass fraction of air from 0 to 1.
The results indicate that variation of the mass fraction of air
has a significant effect on ER. Increasing the mass fraction of
air in the suction flow increases the ER. It is also suggested
that higher pressure in the motive flow decreases the effect of
the variation in the mass fraction of suction flow on the ejector
ER. The sensitivity of ejector performance in terms of the
4697
Nomenclature-----------------------------------------------------------------------CR
dis
E
eff
ER
K
M
mi
m&
t
mot
NXP
P
R
S
Suc
T
t
ij
tot
U
*
i, j,k
: Compression ratio
: Discharge flow
: Total energy
: Turbulent dissipation rate
: Effective
: Entrainment ratio
: Thermal conductivity
: Turbulent kinetic energy
: Mach number
: Molar mass
: Mass flow rate
: Dynamic viscosity
: Turbulent viscosity
: Motive flow
: Nozzle exit position
: Static pressure
: Density
: Refrigerant
: Source term
: Suction flow
: Temperature
: Time
: Stress tensor
: Total
: Velocity
: Mass fraction of ith species in the mixture
: Liquid surface tension
: Critical condition
: Space components
References
[1] D. W. Sun and I. W. Eames, Recent developments in the
design theories and applications of ejectors- a review, J.
Inst. Energ., 68 (1995) 65-79.
[2] Y. K. Kim, D. Y. Lee, H. D. Kim, J. Hahn and K. C. Kim,
An experimental and numerical study on hydrodynamic
characteristicsof horizontal annular type water-air ejector,
4698
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
M. Shafaee et al. / Journal of Mechanical Science and Technology 29 (11) (2015) 4691~4700
4699